To sum this up:<p>A "dark pool" is simply a trading exchange where orders aren't published (trades, of course, are). Huge investment banks run dark pools as a service for clients. The premise behind a dark pool is that institutional buy-side investors would use them instead of a lit exchange because they want to trade against other buy-side investors, without market-makers and aggressive, automated sell-side firms to skim off their profits. Barclays LX is a dark pool set up by Lehman.<p>It turns out that the idea that big buy-side investors can trade with each other without middlemen in a dark pool is probably a fiction. The premise probably doesn't hold. There are at least two big reasons for this:<p>1. Giant institutional investors tend to take similar positions. If one giant firm is gobbling up FCOJ, the others probably want to trade in the same direction. So when you restrict your trading partners to similarly structured and sized firms, you tend not to have counterparties to take the opposing sides of trades.<p>2. The timing of decisionmaking at giant firms is slow, and there aren't all that many of them relative to the market as a whole. So even if there are counterparties to match for trades, that match probably doesn't happen within the window of time that the traders actually want to trade (you want to move a block of stock in minutes or hours, but if you can only trade with other giant mutual funds, it might take days to find a match; think about the difference between filling a market order on ETrade versus trying to sell your house).<p>So for Barclays to make LX actually do something, rather than just acting like a frustratingly inert list of big firms, they needed to get sell-side firms to trade there too. Which they did.<p>The subtext of the Bloomberg post is that getting the sell-side into the LX dark pool was probably a necessity, because when you look at the revenue numbers, Barclays was making its real money from the buy-side; the HFT traders paid it a pittance compared to the mutual funds. All things being equal, it was economically irrational for Barclays to court HFT firms. Of course, in reality it was entirely rational, because without the automated traders, no trades would have happened at all.
Dark pools didn't start off as a bad idea.<p>Their purpose was to allow big institutions to "hide" their block orders. Previously if an institution wanted to trade a big block, it had to break it up via a VWAP or POV algo and trade it over the day or to go to a sell side institution to find a buyer, in which case the broker would take a commission that is often 10x what the regular DMA commission would be.<p>To be fair, this is still what most buy side firms do these days. Dark pools really haven't helped much in this regard as many funds now just bypass them when they can.<p>> Meanwhile, Barclays was advertising LX to high-frequency traders by offering them more information, lower fees, and faster connections than it gave to institutional investors.<p>This is the basis of problem here.<p>A dark pool doesn't publish a bid/ask like other exchanges do. The idea was that everyone would be on equal footing wrt the information present. It appears as though Barclay's has been giving HFTs additional information, including the bid/ask spread in some cases, which totally defeats the purpose of having a dark pool to begin with.<p>The reason they did thsi is that there probably only needs to be one dark pool for the entire US, but each bank realized that it could make some serious money running their own, the only problem is that now you have the same problem as any web 2.0 market place startup does.<p>"How do you get the buyers and sellers"?<p>The HFTs started offering the banks boat loads of money to have access to the order flow in their dark pools if the bank reached certain thresholds of flow. To get to the required flow the banks started routing as many orders as the could, keeping RegNMS in mind, to their dark pools first before settling the orders on other exchanges.<p>I've been told with a straight face by some sell side brokers that I could not tell their SMART order routers to skip dark pools when trying to send my order to the market, again RegNMS excepted. It wasn't until we stopped trading through them that they relented and allowed us to by pass their dark pool.
Let's ignore the fact that Barclays manipulated the data underlying the chart in Figure 1. Think about what the chart is intended to convey.<p>One-second alpha, on the vertical axis, measures an "excess return" certain market participants earned over a 1-second trading interval. It represents skill in trade execution. What Barclays is saying, by pointing giddily at the bubbles in the lower-left quadrant, is that it has collected a fruitful variety of institutions particularly rotten at competently executing their trades. The customer being pitched is being told that it, being smarter than all those clowns, can now take advantage of their naïveté.<p>If I were an existing customer, I'd be insulted to be so portrayed. Or maybe not. Maybe I'd assume I was the lone smart bugger in the top-right. I guess that's what this chart counts on to generate, not lose, business.
From the article:<p><i>The complaint is long on evidence of false advertising, but shorter on evidence that the "predatory trading" was actually predatory. There's a certain amount of foamy fulmination at "predatory trading," but nowhere is there any evidence suggesting that Barclays's institutional clients were actually harmed by it.</i><p>He then goes on to suggest that the lawsuit is therefore essentially frivolous (in his words, "a self-referential argument.")<p>His reasoning is flawed, of course. For one thing, commercial fraud is <i>always</i> harmful, even if you can't always prove that someone suffered specific losses for it. For another, the banks agreed explicitly agreed to comply with all pertinent legislation (including the Martin Act) in exchange for the privilege of obtaining a charter which allowed them to become banks in the first place. Whether one thinks the Martin Act is overly broad or gives too much leeway to prosecutors is of secondary importance.
I don't know enough to have an opinion on HFTs and whether they actually provide liquidity. However, creating sub-markets that are illiquid if you remove them, but become liquid if you add them back in, strikes me as a pretty powerful bit of evidence in HFT's favor. Rebuttals? (Serious question.)
Matt Levine is awesome. Thanks for sharing. He's also the world kingpin of footnotes. It's almost as if he's a frustrated academic who needs to hide his humor in citations.
I have a couple of questions for the people on here who actually understand all this stuff.<p>The argument, as I understand it, for why it is useful to have HFTs in markets like the one in the article, is that they provide extra liquidity. My understanding of liquidity is that it is some kind of measure of how quickly an asset can be sold. I'm sure there are some formal definitions, but that has to be something like [asset value/sale time], where larger is more liquid.<p>It seems to me, though, that liquidity is necessarily very time-scale dependent. For example, there is basically no liquidity in any market at the picosecond (or whatever is faster than current trading) timescale, because (by definition) no-one trades that fast. So really, although liquidity can be calculated as a rate, it is more appropriate to look at the distribution of available liquidity over different time scales. In other words, being able to sell $10 of assets in 10 days is not really the same as being able to sell $1 of assets every day for 10 days.<p>My questions are:<p>1) is my understanding of liquidity as being timescale dependent correct, and if not, why not?<p>2) if yes, what is the social value of "faster" liquidity?<p>3) if "faster" liquidity has greater social value, is there any point at which the cost of maintaining that liquidity (which is what I understand HFT profits to be) exceeds the value of having liquidity at that timescale?<p>edit: spacing for visual clarity
I struggle to feel bad for these institutional investors. These investors are mostly fund managers who make on the order of 1% of assets under management yearly... Arguably, their jobs are composed of two functions: decide on investment decisions, and execute on those decisions. Yet - they are unable to do their own analysis of execution quality.<p>A simple determination of execution quality does not require a PhD in math - it requires comparing the price of execution to the market shortly after the execution occured (If it moved against you, you could argue there was "adverse selection"). I am not saying what Barclay's did is right. I just feel it says more about the inept behavior of institutional investors that they were not able to smell a rat earlier on.<p>The article rightly points out that HFT drives dark pools. If an institutional investor didn't know that, I am very confused why we pay them so much money...
I dug up the original chart, before Tradebot was removed: <a href="http://jackgavigan.com/2014/06/30/barclays-smoking-chart/" rel="nofollow">http://jackgavigan.com/2014/06/30/barclays-smoking-chart/</a>
There seems to be a false dichotomy here - free-for-all or no automated traders. Automated firms are included in the diagram, so Barclays wasn't hiding them. It was, however, claiming to remove those it considered bad actors, and then wasn't actually doing so. Surely allowing HFT firms but removing those (specifically, Tradebot) that didn't follow the rules would still have left enough trade flow to enable the buy-side to get value? If it's bad enough to warrant removal from the diagram, it's bad enough to be removed from the pool, leaving every other participant to continue as before.
Amazing info and article.<p>In the end, Barclays created the LX dark pool to get more fees from individual investors and smaller investors. They used HFT firms as the carrot and stick to extract more revenue from the suckers using HFT and the volatility/liquidity that it can provide.<p>> Meanwhile, Barclays was advertising LX to high-frequency traders by offering them more information, lower fees, and faster connections than it gave to institutional investors.<p>Seen from an investor who bought the marketing, this is dishonest. But seen from the Barclay's boardroom, if they make all their money off of non HFTs, they need to increase buying/selling on their smalls and institutional investors. So of course Barclays would make it really cheap for the HFTs, all the more action to extract fees and commissions from the clients/suckers that pay.
Are there any services out there that allow automated trading, but as a game? There are many services that let you do manual trades as a game, but I've never seen automated ones.
Some thoughts on this as the former head of electronic trading product management at Deutsche Bank<p><a href="http://dave-hunter.com/why-barclays-lied" rel="nofollow">http://dave-hunter.com/why-barclays-lied</a>
Is this not a clear case of Fraud & Mis-representation?
If a person had pulled this kind of stunt you would see criminal chrges flying all over the place ..Why are bank executives at Barclays not going to jail over this?
"Meanwhile, Barclays was advertising LX to high-frequency traders by offering them more information, lower fees, and faster connections than it gave to institutional investors."<p>Whether intentional or not, it sounds like they built a combination public swimming pool / alligator farm.<p>Marc Andressen recently penned a screed that made the rounds about how Sarbox and friends have killed the IPO. He's not wrong there. IPOs are happening much later and less frequently because of regulation.<p>Where he's wrong is... well... he's basically like a resident in a high crime neighborhood asking why all the shops have bars on their windows. It's driving away business after all! Let's take those bars off the windows...<p>But as soon as you do that, you get a smash and grab the next night. Sorry Marc, but there's bars on the windows because the neighborhood is full of thugs.<p>The ultimate problem -- and the ultimate thing that has killed the IPO -- is that the financial system is full of glorified white collar street hustlers who look at markets as something to bust out for short term profits.<p>Like the proverbial bad neighborhood I think the only solution might be to move out. I can see technologies like Bitcoin and systems like crowd funding as the embryonic beginnings of a new financial system with greater intrinsic transparency and with cryptographic authentication baked in from the get-go. Perhaps, with that added transparency, it'll be possible to engineer in a fundamentally better security model to discourage predation.